> High!
>
> To make testing easier, I resorted to a simple spherical "test planet"
> and tried various camera positions and illumination angles with all five
> scattering types (and yes, it wasn't media but the isosurface which
> slowed down my renderings - now each scene took about 70 seconds!).
>
> The simplified test planet is a sphere with a radius of 5,100 POV units
> (1 unit equals 1 kilometre); the media is contained by a sphere with a
> radius of 5,150 units. In all scenes, the illumnating star is a simple
> point light; in all ground scenes (all scenes from row 3 on), the camera
> is placed 0.0017 units above the planet's surface.
>
> Method, samples and intervals settings are default (see code attached
> here).
>
> The most striking flaw with all scattering types is the extremely dark
> sky at low illumination angles - the sunrise looks like on Mars but not
> like on a planet with Earth-like atmospheric pressure. Could using an
> area light instead of a point light be helpful here? Or do I have to
> fake the indirectly illuminated sky before dawn by using an additional
> (shadowless) fill light source?
>
> Reactor's suggestion of adding an absorbing component did not work at
> all - regardless which ratios between scattering and absorbing, the sky
> turned even darker!
>
> Then, at medium illumination angels (~45 degrees), the otherwise pretty
> realistic sky with scattering type 1 and 4 still turns too greenish
> close to the horizon...
>
> See you in Khyberspace!
>
> Yadgar
>

> density
> {
> spherical
> density_map
> {
> [0 rgb <0.25, 0.5, 1>]
> [0.005 rgb <0.25, 0.5, 1>]
> [0.011 rgb <0.9, 1, 1>]
> [1 rgb <0.9, 1, 1>]
> }
> }
> }
> }
> }
> }
>

Your spherical pattern have a value of rgb<0.25, 0.5, 1> EVERYWHERE 
outside a radius of 1. That means that the part that you can see only 
have that value, as the surface is at a radius of 5140.

You may try scaling the media to the radius of it's container.
Add:
scale 5190
just after the density_map.


Alain

Post a reply to this message